someone said u cant OC the card, but Ntune seems to do it.
I had it at 750core/900mem/1700 shader, some higher ones would crash game. (But thats not probem). Nvidia system monitor shows the OC'd values all full when game splash starts, etc...then when entering game just the mem OC trails down to 300 and stays there. the other OC values stay up. It does this if I use default clocks also (600/800/1500). I guess I can try rivatuner but want to understand this first if possible. Anybody ever see anything like this or have any theories?![]()
-
-
im having the same problem even in game play it clocks down and then goes back up idk whats going on....
-
run the clocks 698/988/1798..and you'll get great results, and no stability loss when gaming..and RivaTuner sucks with the 9800m GTS. Use Ntune, download link is found in the Nvidia Driver thread at the top of this forum.
-
i only used riva tuner to monitor my clocks to make sure they were staying the same i overclocked the gpu with nvidia system tool it worked great but my clocks would throttle down on me...
-
you have no need for rivatuner..and if you're watching your clocks change..then you're not gaming. The higher clocks you're setting are the 3d performance clocks which kick in while GPU(vid card not processor) intense processes such as games, are running. The GPU would burn out if it was locked into the high clocks all the time. That's the same reason anyone who uses RMClock to undervolt the CPU(processor, not vid card) would never lock it in High Performance mode.. it is a speed stepping CPU and you use Performance On Demand mode cause like the vid card..it can change clocks as necessary to provide for the demand of a process.
-
E-wrecked- Im specifically talking about seeing just the mem clock go down while playing ETQW in non-fullscreen where I can see the nvidia system monitor at the edge of my desktop. The core and shaders clocks stay OC'd. It doesnt seem to do it in Enemy Territory, just Quake Wars. (and I dont know what other games). yes we are talking about in-game. I know about the
auto-throttle in 2d etc-thats not the issue. it doesnt do it for the first serveral seconds in game...then it puts mem at 300 and stays there. I made sure laptop was in performance power mode. It's reading temp on GPU at 52-53C. -
looks like Im using Nvidia Systems Tools 6.03, which includes
Nvidia Performance Group that does the OC. I thought this was
Ntune? Maybe I should get rid of this and try to get Ntune proper? My desktop doesnt do it in XP. But then it doesnt auto-throttle either, it stays at full speed (no OC these days, 8800 gt 512). I doubt it has to do with Vista in general (though that does cause problems running ET-ask me for answers). I'd bet $5 it has something to do with the throttle function on the lappy. It does this on QW but not ET. I wonder what other titles it underclocks memory on? Would everyone please try some other titles in non-fullscreen and run Nvidia system monitor mini-window on the side of the desktop? thanks! -
P.S. you can edit your posts, you don't have to quadruple post...
-
can you explain a bit more exactly where you are looking that you see these memory clocks drop? are you looking at "GPU Memory Clk" in Nvidia system monitor or are you looking elsewhere?
-
u can delete ur own posts. click on edit, then delete, then turn on the 'delete this post' button. u dont have to give reason. notice my post that says to delete yours is gone? Yes I see the gpu mem clock on Nvidia system monitor.
i just put them (gpu temp,core clk,mem clk, shader clk)on the edge of my desktop, then run the games in non-fullscreen. I feel the drop is real because I dont see it when running ET, just QW.
IM THINKING THIS IS POWERMIZER ISSUE, LOOKING FOR BEST SOLUTION...RESOLVED...USE 'POWERMIZER OFF' inf . There is a package 'Vista64_180.43' compressed size 47.5 mb. Its one of the DLs from E-wrecked's main thread with driver upgrade links.
You have to look for right one and get this one specifically. There is a larger on I used before that caused problem. Anyway, inside the 'right' package is a folder 'ALT infs' if you are having my problem use the one in folder named 'Powermized off' (it even says that in the read me). this isnt SLT but that sure was the solution! So yeah no wacky reg edits or anything. just need the right nv_disp file. -
If you do a search there is a walk through on this forum of how to disable powermiezer.
Also Ntune is the Nvidia Performance Monitor thing that you get when you install that portion of the System Tools. It's the only way to do it, there is no Ntune proper anymore, and Rivatuner doesn't work with these Laptops atm (or at least didn't). I also endorse E-Wrecked's clocks but I bumped them up to the even numbers lol.
ALSO.. and this may have changed, System Tools 6.02 doesn't work with the 180+ drivers very well at all (lockups), but there is a new 6.03 version out. Not sure if 6.03 will work with drivers older then 180.xx, not saying it won't, just saying I haven't tried it.
I personally still use the Xfasted 178.85 drivers so I'm sticking with the 6.02 version of Ntune. My lappy is currently working amazing and I'm of the 'If it ain't broke, don't fix it' school of thought, at least until I buy my 2nd HDD and go RAID 0 and have to blow everything out anyway.
Just be aware if you run into problems that these could be the issues. -
The old ntune propper may bluescreen Vista64, at least, it did to me. I've been using 6.03 with DOX's 178.24 driver with no problem and nice results. E-Wrecked's numbers have worked for some on this forum but may not work for others. Overclocking is a bit of a dark art and requires a lot of trial and error to get the last ounce of performance out of your GPU. E-Wrecked pushes the limits and his numbers may still cause you problems. OTOH, you may find you can stably go beyon what he's done... if you dare...
-
sys tools 6.03 is what Im using and its working fine. i think the problem is in the inf (nv_disp). Im gonna keep trying different ones and maybe Ill get it right. Right now I have the 'powermizer off' one running and its great for games but doesnt clock down for 2d on its own and keeps the fan on medium, even when I manually clock down and there no load. The factory drivers/inf slowed the fans down and even stopped them, but would also do this annoying pulsing of the fan speed when it was running. I would rather it just run slow continous or something. did you have that fan pulsing? correct it?
-
I have powermizer on and my memory clks stay put, but I'm not using dox's customized so maybe dox edited some powermizer configuration.
-
I didnt think Xfst drivers had powermizer.. I searched the reg after installing, and never found it. I mean, there is the one instance of powermizer with the 3333 setting.. I changed it to 3322, and heck.. even before changin never had a problem. Search the forums for "disable powermizer" and the first link that results should be the correct walktru..takes 2.5 minutes.
-
Here's my 'final' solution: I had been trying 180.43 all this time. I did not want to start messing with reg. So I started trying 177.41. Xfst wouldn't work at ALL for me. What finally worked was something skatehouse posted ( forum.notebookreview.com/showthread.php?t=265740 )with separted inf and driver dls. inf: rapidshare.com/files/125261957/nv_disp.inf.html driver: vaguetech.com/forum/downloads.php?do=file&id=252 (sorry dont know how to make links). Anyway, I pretty happy with it. Dont know if 177 is any slower, but will take it so powermizer will work. have my OC at 733/972/1900 63-70C stable in Quake Wars. that's a good OC huh? I may try a little more since the fan seems to respond right to higher temps. sry had to truncate urls due to server restricting me
-
-
I finally RAIDed and reinstalled Vista64 last night with two Seagate 320GB Momentus 7200.3. This gave me the opportunity to do a fresh install of DOX's modded 178.24 driver (I downloaded it a week or so ago). Sure enough my clocks are pegged at 600/799/1500 even when just looking at the desktop. What's odd, is that when I first installed DOX's drivers on the preloaded OS, I had followed the instructions to remove the old Nvidia drivers, including booting in safe mode and running driver sweeper prior to installing the DOX modded 178.24 driver and yet my GPU clocks would downthrottle when I wasn't gaming or benchmarking (I got a nice 400 point bump in 3DMark just from the driver). It's an interesting observation. It could imply that it's more than just a driver setting that governs gpu throtling. Or it could imply that I didn't get all the nvidia drivers off my preloadd OS before the install and some remnant preserved and it had a nice side effect. I have not seen the gpu downthrottle during gaming.
-
Down throttle of GPU when not stressed is a desirable thing on a loptop for temp. purposes. Really it is always a good thing.
-
I did a little research and discovered that the way to get it to downclock when not in 3D mode was to go into regedit and changing the PerfLevelSrc found under Computer\HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Video\{?????????}\0000
In my case, it was 2222 and I changed it to 3333. Setting it to 3322 allows it to downclock only when on battery.
It made no difference in 3DMark scores.
Another thing that happened to me for the first time in a while, was that I got TZS0 to hit 98C for an extended period of time running Furmark and playing chess at the same time. However, it do not downclock the GPU and did not appear to affect the Furmark framerate. Once I killed Furmark, TZS0 immediately dropped to 41C. I'm inclined to agree with E-Wrecked on the source of the problem and at least with the P8400, it does NOT cause a downthrottle of the cpu. So, I'm perfectly happy to just ignore it.
http://forum.notebookreview.com/attachment.php?attachmentid=26207&stc=1&d=1225429811Attached Files:
-
-
Well, well, not sure what to make of this... I tried Daedric's GPU OC values 720/1005/1860 and ran 3DMark06. Very shortly into the Firefly forrest segment, I noticed the framerate drop to 9-11FPS. Since I had CPUID HWMoitor up, in the screen flash between Firefly Forrest and CPU Test 1, I could see that TZS0 had spiked at 98C, So I know that the TZS0 spike happened about the same time the framerate dropped in Firefly Forrest and lasted into CPU Test 1. However, watching the CPU test, it didn't seem to be affected. I had done some prior runs with no overclocking, 50MHz OC on all GPU clocks and 100MHz OC on all GPU clocks. Comparing the CPU results I saw CPU scores of 2036, 2044 and 2032 on the three original runs and 2021 on the run where I saw the Firefly framerate drop to 9-11FPS. That is less than a percent drop and is not statistically significant. The SMS2 scores for my first three runs were 4123, 4274 and 4386 respectively. In the run where I saw the framerate drop, the SMS2 score was 3983, significantly lower than the SMS2 score in even the baseline no OC run. So, the question then becomes, did the TZS0 spike cause a CPU throttle or a GPU throttle? Since I saw no significant drop in the CPU score, I was kind of leaning toward a CPU throttle, even though HWMonitor showed the max GPU temp at 65C. But to prove it, I needed history, so I fired up RivaTuner to plot the GPU clock history and ran 3DMark06 again. In this run, the framerate drop occurred later in Firefly and seemed to last into CPU test 1 and this was borne out by a SMS2 score of 3983, almost 270 points higher than before, but still below the baseline score of 4123. also this tiem, the CP score was 1835, significantly below the baseline score of 2036. Since I've been told the GPU has little or no effect on the CPU test scores, this tends to make one think the CPU was throttled and not the GPU. And the graphs from Rivatuner showed no drop in GPU clocks, so this, along with the CPU scores, leads one to believe that TZS0 caused a CPU throttle. It is strange that when I am not overclocking the GPU and I see TZS0 spike, I don't see a CPU throttle.
p-7811fx GPU mem clock throttles down stuck, heyelp pls!
Discussion in 'Gateway and eMachines' started by el_touristo_duo, Oct 27, 2008.